Qualitative age interactions (or effect modification) suggest divergent pathways for early-onset and late-onset breast cancers

Abstract

21036 Background: Notwithstanding some recent declines, breast cancer incidence rates have risen for decades, though not equally for all age groups. We used the National Cancer Institute's SEER program to further explore the effect of aging upon breast cancer incidence. Materials and Methods: The SEER program collected data on n=494,543 in-situ + invasive female breast cancer cases, newly diagnosed during 1974–2003. Temporal trends by race, stage, and grade were stratified by age at diagnosis in decades: 20–29 to 80+ years. Results: We observed age interactions over time. For example, as the specification of grade improved from 1974–2003, temporal trends for high and low grade tumors varied with age. Among women ages <40 years, high grade lesions were more common than low grade tumors for all time periods. Among women ages 40+ years, high grade lesions were more common during the early years, and then rates crossed, after which low-grade tumors were more common than high grade lesions. Conclusion: Age at diagnosis was both a quantitative (non- crossover) and qualitative (crossover) effect modifier. The crossing of rates from high to low grade tumors among women ages 40+ years in the 1980s is consistent with more aggressive breast cancer screening, with mammography preferentially detecting low grade tumors among women targeted for screening, i.e., ages 40–80 years. Though once thought to be rare or artifactual, qualitative interactions or effect modification suggest etiologic heterogeneity in an otherwise homogeneous disease process. Indeed, if true, qualitative age interactions imply divergent pathways for early-onset and late-onset breast cancers. No significant financial relationships to disclose.